The Slow and Long-Lasting Blockade of Transporters in Human Brain Induced by the New Radafaxine Predict Poor Reinforcing Effects Nora D. Volkow, Gene-Jack Wang, Joanna S. Fowler, Susan Learned-Coughlin, Julia Yang, Jean Logan, David Schlyer, John S. Gatley, Christopher Wong, Wei Zhu, Naomi Pappas, Michael Schueller, Millard Jayne, Pauline Carter, Donald Warner, Yu-Shin Ding, Colleen Shea, and Youwen Xu Background: (2S,3S)-2-(3-Chlorophenyl)-3,5,5,-trimethyl-2-morpholinol hydrochloride (radafaxine) is a new antidepressant that blocks dopamine transporters (DAT). A concern with that block (DAT) is their potential reinforcing effects and abuse liability. Using positron emission tomography (PET) we have shown that for DAT-blocking drugs to produce reinforcing effects they must induce Ͼ50% DAT blockade and the blockade has to be fast (within 15 minutes). This study measures the potency and kinetics for DAT blockade by radafaxine in human brain. Methods: PET and [ 11C] were used to estimate DAT blockade at 1, 4, 8, and 24 hours after radafaxine (40 mg p.o) in 8 controls. Plasma pharmacokinetics and behavioral and cardiovascular effects were measured in parallel. Results: DAT blockade by radafaxine was slow, and at 1 hour, it was 11%. Peak blockade occurred at about 4 hours and was 22%. Blockade was long lasting: at 8 hours 17%, and at 24 hours 15%. Peak plasma concentration occurred about 4 to 8 hours. No behavioral or cardiovascular effects were observed. Conclusions: The relatively low potency of radafaxine in blocking DAT and its slow blockade suggests that it is unlikely to have reinforcing effects. This is consistent with preclinical studies showing no self-administration. This is the first utilization of PET to predict abuse liability of a new antidepressant in humans based on DAT occupancy and pharmacokinetics.

Key Words: PET, kinetics, abuse, reinforcement, striatum prevent people from taking medication regularly. Radafaxine is the (ϩ) isomer of (Sanchez and Hyttel 1999). epression is one of the most prevalent psychiatric disor- Hydroxybupropion is chiral, and the ability to block the DAT and NERT resides in the (ϩ) isomer, radafaxine (Sanchez and Hyttel ders of adulthood (Kessler et al 1994). The lifetime ϩ prevalence for major depression in the United States in 1999). The ( ) isomer of hydroxybupropion also seems to D underlie the efficacy of hydroxybupropion in a mouse model of persons 15–54 years of age has been estimated to be 17.1% (Blazer et al 1994). Although there are a wide variety of depression (forced swimming) (Damaj et al 2004). In addition, ␣ ␤ antidepressant medications, there is still need for newer medica- this isomer is also an antagonist of the 4 2 receptor tions. There are still a significant number of depressed patients and has been shown to interfere with the acute effects of nicotine who do not respond or have inadequate responses; the overall in mice (Damaj et al 2004). mean response rate for was recently estimated to A concern regarding drugs that block DAT is that they might be 62% (Lam et al 2002). It would also be desirable to develop be reinforcing (Ritz et al 1987). Indeed, human imaging studies new antidepressant drugs with a better safety margin and fewer have shown a linear relationship between the degree of DAT side effects than those currently available. blockade by drugs such as cocaine and (MP) In the study reported here, we evaluated a new antidepres- and their reinforcing effects when they are injected intravenously sant drug (2S,3S)-2-(3-Chlorophenyl)-3,5,5,-trimethyl-2-morpho- (IV), as assessed by self-reports of drug effects such as “high” and linol hydrochloride (radafaxine; GlaxoSmithKline, Research Tri- “rush” (Volkow et al 1997, 1999b). Using positron emission angle Park, North Carolina). The efficacy of radafaxine in tomography (PET), we have assessed the properties of DAT- depression is attributed to its activity at both the dopamine blocking drugs that are associated with their reinforcing effects. transporters (DAT) and the transporters (NERT) We have shown that for these drugs to be reinforcing, they have (Sanchez and Hyttel 1999). It is thought that radafaxine might to block more than 50% of the DAT within a relatively short Ͻ have advantages over other antidepressants in terms of reduced period ( 15 min from administration) and clear the brain rapidly side effects, such as diarrhea and nausea, which very often to enable fast repeated administration (Volkow et al 1995a, 1997, 1998). In this study, we used PET to evaluate the effects of radafax- From the Brookhaven National Laboratory (NDV, G-JW, JSF, JY, JL, DS, JSG, CW, NP, MS, MJ, PC, DW, Y-SD, CS, YX), Upton; Departments of Psychiatry ine at the DAT in the human brain (potency and pharmacoki- (NDV) and Mathematics (WZ), State University of New York at Stony netics) as a means to predict its reinforcing effects and hence its Brook, Stony Brook, New York; and GlaxoSmithKline (SL-C), Research abuse liability. blockade was measured 11 Triangle Park, North Carolina. with the DAT radioligand [ C]cocaine in eight healthy control Address reprint requests to Nora D. Volkow, M.D., Brookhaven National subjects. To assess the pharmacokinetics of DAT blockade, we Laboratory, Medical Department, Upton, New York 11973; E-mail: measured DAT blockade at different times after radafaxine [email protected]. administration (1, 4, 8, and 24 hours) in each individual, in Received July 27, 2004; revised November 29, 2004; accepted December 4, addition to a baseline [11C]cocaine scan before drug administra- 2004. tion. To assess the level of DAT occupancy by radafaxine, we

0006-3223/05/$30.00 BIOL PSYCHIATRY 2005;57:640–646 doi:10.1016/j.biopsych.2004.12.007 © 2005 Society of Biological Psychiatry N.D. Volkow et al BIOL PSYCHIATRY 2005;57:640–646 641 selected a dose of 40 mg, which is the targeted dose for the late distribution volume ratios in caudate, putamen, and ventral treatment of depression. The behavioral and cardiovascular striatum, with a graphic analysis technique for reversible systems responses to radafaxine were monitored during the PET studies. (Logan et al 1990), modified to avoid arterial sampling (Logan et al 1996). The distribution volume ratio, which is the ratio of the Methods and Materials distribution volume (DV) in striatum to that in cerebellum

(DVSTR/DVCB) and which corresponds to the binding potential Subjects Bmax/Kd ϩ1, was used as an estimate of DAT availability (Logan Ϯ Ϯ Eight healthy male subjects, 31 6 years of age (mean SD), et al 1994, 1996). Dopamine transporter occupancies were were studied. Written informed consent was obtained from all Ϫ calculated as [(Bmax/Kdbaseline Bmax/Kdradafaxine)/Bmax/ subjects after a complete description of the study and following ϫ Kdbaseline] 100. the guidelines set by the institutional review board at To determine whether radafaxine significantly induced DAT Brookhaven National Laboratory (Upton, New York). blockade, we performed one-way repeated-measures analysis of variance (ANOVA), comparing the Bmax/Kd at different times. PET Scan ϩ Upon rejection of the ANOVA hypothesis of equal DAT blockade The PET studies were carried out with an HR tomograph across time, we used the Dunnett method for comparison with a (resolution 4.5 ϫ 4.5 ϫ 4.5 mm full width half-maximum, 63 11 control to compare the Bmax/Kd obtained at different times after slices), with [ C]cocaine used as a DAT ligand (Fowler et al radafaxine administration with the baseline, controlling the ex- 1989). Methods for positioning of subjects, catheterizations, perimentwise error rate to be .05 (two-sided). To assess whether transmission scans, and blood sampling and analysis have been the levels of DAT occupancy differed for the different time points published (Fowler et al 1989). Briefly, emission scans were after radafaxine administration, we performed repeated-mea- started immediately after injection of 4–8 mCi of [11C]cocaine Ͼ ␮ sures ANOVA, and post hoc pairwise comparisons were then (specific activity .2 Ci/ mol at time of injection). A series of 20 done with the Tukey method, controlling the experimentwise emission scans were obtained from time of injection to 54 min. 11 error rate at .05 (two-sided) to assess at which time points the Subjects were scanned five times with [ C]cocaine over a DAT occupancies differed. Stepwise regression and the least 2-day period; the first scan was done with no intervention and squares method were used to fit the temporal curves of plasma was used as a baseline, and the other four scans were performed drug concentration and DAT occupancy and to find the paramet- 1 hour, 4 hours, 8 hours, and 24 hours after a single oral ric relationship between plasma radafaxine concentration and administration of 40 mg of radafaxine. Venous blood was drawn DAT occupancy in striatum. Spearman’s rank correlation analysis for quantification of plasma concentrations of radafaxine before was used to further examine the relationship between DAT and at .5, 1, 1.5, 2, 3, 3.5, 4, 4.5, 5, 6, 7, 8, 8.5, 9, 12, 24, 24.5, and occupancy and the concentration of radafaxine in plasma. For 25 hours after administration of radafaxine. Human plasma the comparison of cardiovascular measures, we averaged the samples were assayed for radafaxine with a method based on measures taken just before and during the first 30 min of each protein precipitation with acetonitrile, followed by liquid chro- scan. For the comparisons of the descriptor of drug effects, we matography/mass spectometry/mass spectometry (liquid chro- also averaged the measures obtained before and during each matography with tandem mass spectrometry; LC/MS/MS) analy- PET scan (three ratings). Paired-samples t-tests were used to sis with positive-ion Turbo IonSpray ionization (TISI) (lower limit ␮ assess whether there were differences in the cardiovascular and of quantification, .5 ng/mL for a 100- L aliquot of human behavioral measures from the baseline scores. The normality plasma). assumption was examined with the Wilk-Shapiro test. Behavioral Measures To simulate an analogue rating scale, participants were in- Results structed to orally respond to each descriptor with a whole number between 1 and 10 for the self-report of “high,” “mood,” Radafaxine significantly blocked DAT, as assessed by signifi- “restlessness,” “anxiety,” “tired,” “alertness,” “desire for more cant differences in the Bmax/Kd measures in caudate [F(4,28) ϭ drug,” and “control over intake if exposed to more drug.” Ratings 8.7; p ϭ .0001], putamen [F(4,28) ϭ 13.7; p Ͻ .0001], and ventral were obtained 5, 27, and 47 min before each PET scan evaluation striatum [F (4,28) ϭ 6.6; p ϭ .0007] (Figure 1, Table 1). Post hoc for the baseline scan, then 1, 4, 8, and 24 hours after radafaxine Dunnett test at the experimentwise error rate of .05 showed that scans, as previously described (Wang et al 1997). Heart rate and with respect to baseline the differences were significant for all blood pressure were monitored continuously. but the 1-hour measure in caudate and ventral striatum and were significantly different for all time points in putamen. Analyses Peak DAT blockade did not occur until 4 hours after radafax- Emission data were corrected for attenuation and recon- ine administration, at which time blockade corresponded in structed with filtered back projection. For the purpose of region caudate to 22% Ϯ 5% (range, 15%–30%), in putamen to 20% Ϯ identification for the [11C]cocaine scans, time frames from dy- 8% (10%–30%), and in ventral striatum to 20% Ϯ 8% (10%–33%) namic images taken at 10–54 min were summed and the (Figure 2). The maximal occupancy achieved for any one of the summed image resliced along the anterior commissure–posterior regions was 30%–33%. At the .05 significance level, DAT occu- commissure line. Planes were added in groups of 2 to obtain 12 pancy was found unequal across time in caudate and putamen, planes encompassing the caudate, putamen, ventral striatum, but this difference did not reach significance in ventral striatum and the cerebellum for region of interest placement. The cau- (p ϭ .13). Post hoc Tukey pairwise comparison at the .05 level date, putamen, ventral striatum, and cerebellum were measured showed that in caudate and putamen, differences in DAT occu- on 4, 3, 1, and 2 planes, respectively, and right and left regions pancy were significant between 1 and 4 hours but not between were averaged. These regions were then projected to the dy- the other hours. Repeated-measures ANOVA to assess whether namic scans to obtain concentrations of 11C versus time. These there were differences in DAT occupancy in the three striatal time activity curves for tissue concentration were used to calcu- regions showed no significant differences for any of the time

www.elsevier.com/locate/biopsych 642 BIOL PSYCHIATRY 2005;57:640–646 N.D. Volkow et al

Figure 1. Distribution volume images of [11C]cocaine for one of the subjects at baseline and at 1, 4, 8, and 24 hours after administration of 40 mg of Figure 2. Dopamine transporter (DAT) occupancy by radafaxine at different radafaxine. times after its administration. Occupancy was significantly lower at 1 than at 4 hours after its administration for all striatal regions and also between 1 and 24 hours in caudate only. Values correspond to means and SDs. points at the significance level of .05, which indicates that radafaxine has no special selectivity within the striatum. Because (Figure 4). The fit has a low coefficient of determination at R2 ϭ of this correlation measure with plasma, concentrations were .26 because only four different time points were available. Also, performed only for the whole striatum (weighted average of its statistical reliability is limited by the fact that the measures are caudate, putamen, and ventral striatum). not independent from one another because a given subject The pharmacokinetics of radafaxine are shown in Figure 3. contributed to several time points. According to this curve, the The peak plasma concentration of radafaxine occurred between peak DAT occupancy is approximately 19.1%, which occurred hours 6 and 5, which corresponds well with the time when peak between 4 and 6 hours after drug intake. Stepwise least squares DAT blockade was observed. A good fit (R ϭ .84) for the plasma regression revealed a linear relationship between DAT occu- concentration of radafaxine was found by the least squares pancy in striatum and plasma radafaxine concentration. The method, with the estimated plasma concentration at time t being estimated DAT occupancy when plasma concentration is x 104.8 Ϫ 2.3t Ϫ 112eϪt Ϫ 91teϪt (Figure 3) a mixture of linear, corresponds to 5.4 ϩ .14x (Figure 5). The coefficient of determi- inverse exponential, and gamma components. e is a constant and nation for the regression is only R2 ϭ .24, owing to the large the base of the natural logarithm (e ϭ 2.71828). This fit has a variation in the data. The three regression curves, however, agree coefficient of determination of .71, indicating that 71% of the well with each other in the prediction of peak DAT occupancy variation in the data could be explained by this regression. based on peak plasma radafaxine concentration. At the estimated Interpolation according to this curve showed that the peak peak plasma level of 89.7 ng/mL, expected to have occurred 5.4 plasma concentration occurred at approximately 5.4 hours after hours after drug intake, the estimated DAT occupancy is 19.0% drug intake, with an estimated value of 89.7 ng/mL. The temporal according to the above linear relationship between plasma curve for DAT occupancy in the entire striatum was found to be concentration and DAT occupancy. This is in agreement with a combination of linear and inverse exponential terms, with the what was estimated on the basis of the DAT occupancy curve DAT occupancy at time t being estimated as 21.0 Ϫ .3t Ϫ 29.9eϪt showing 19.1% blockade occurring 4–6 hours after drug intake. In addition, we calculated the Spearman correlation to further Table 1. Bmax/Kd Measures in Caudate, Putamen, and Ventral Striatum Before and at Different Times After Administration of Radafaxine

Bmax/Kd Ventral Time Caudate Putamen Striatum

Baseline .82 Ϯ .14 1.03 Ϯ .17 .87 Ϯ .14 1 h .75 Ϯ .11 .91 Ϯ .14 .80 Ϯ .16 4 h .64 Ϯ .08 .81 Ϯ .10 .70 Ϯ .15 8 h .67 Ϯ .09 .85 Ϯ .10 .75 Ϯ .09 24 h .72 Ϯ .13 .88 Ϯ .13 .75 Ϯ .09 Data are presented as Mean Ϯ SD. Repeated-measures analysis of vari- ance corresponded in caudate to F(4,28) ϭ 8.7, p ϭ .0001; in putamen to F(4,28) ϭ 13.7, p Ͻ .0001; and in ventral striatum to F(4,28) ϭ 6.6, p ϭ .0007. Post hoc Dunnett test showed that with respect to baseline the differences were significant at the experimentwise error rate of .05 (two-sided) for all time points in caudate, putamen, and ventral striatum. Bmax/Kd refers to the ratio of the free receptor concentration and the equilibrium dissociation constant. Figure 3. Temporal curve for the concentration of radafaxine in plasma. www.elsevier.com/locate/biopsych N.D. Volkow et al BIOL PSYCHIATRY 2005;57:640–646 643

Table 2. Spearman Rank Correlation Between Plasma Radafaxine Concentration and DAT Occupancy With and Without Time Lag

Striatum Spearman No. of Correlation P Observations

No Lag (1/4/8/24) .53 .002 31 No Lag (1/4/8) .54 .008 23 Lag (.5/3.5/7) .54 .010 22 Lag (.5/3/7) .47 .026 22 Numbers in parentheses are time points in hours. Plasma concentrations at hours .5, 3, and 7 were used. DAT, dopamine transporter.

This study also shows that radafaxine exhibits slow brain pharmacokinetics (slow rate of DAT blockade, long-lasting blockade, and slow clearance from the DAT). In fact, at 1 hour after its administration, DAT blockade was only 10% and was Figure 4. Temporal curve for dopamine transporter (DAT) occupancy in the significantly lower than the peak blockade, which was recorded entire striatum. at 4 hours after its administration. This slow DAT blockade also contrasts with that observed after oral MP, which 1 hour after its examine the monotone relationship between the plasma concen- administration blocks on average 60% of the DAT. tration and DAT occupancy on the whole striatum with and The DAT blockade after radafaxine was long lasting, and without time lag (Table 2). Using all available temporal measures there was still significant blockade at 24 hours after its adminis- (hours 1, 4, 8, and 24), we found a significant Spearman rank tration. Moreover, DAT blockade at 24 hours in putamen and correlation between plasma concentration and DAT occupancy ϭ ϭ ventral striatum did not differ significantly from the levels (r .51, p .003) and no evidence of a time lag effect of plasma observed at peak (4 hours). Indeed, the blockade of DAT by radafaxine concentration toward DAT occupancy. radafaxine was very stable, with a decrease from peak at 4 hours None of the behavioral (Figure 6) or cardiovascular effects of 20%–22% to 12%–16% at 24 hours. (Figure 7) of radafaxine were significant. The relatively low potency of radafaxine at blocking DAT, its Discussion This study shows that radafaxine has a relatively low potency for blocking the DAT and exhibits slow kinetics in the human brain. At the proposed therapeutic dose, it blocked an average of 20%–22% of the DAT after a single dose. Moreover, the maximal blockade in any one subject was 33%. The low levels of DAT blockade achieved with radafaxine contrast markedly with the levels of DAT blockade induced by the DAT blocker methyl- phenidate (MP), which at the oral doses used therapeutically occupies on average 60% of the DAT in the human brain (Volkow et al 1998).

Figure 5. Stepwise least squares regression between dopamine transporter Figure 6. Behavioral measures (self-ratings 1–10) before and at different (DAT) occupancy in striatum and plasma radafaxine concentration (R ϭ .49; times after administration of radafaxine. Values correspond to means and p Ͻ .005). SDs. None of the effects were significant.

www.elsevier.com/locate/biopsych 644 BIOL PSYCHIATRY 2005;57:640–646 N.D. Volkow et al

an explanation for the differences in the reinforcing effects of drugs of abuse as a function of route of administration (Verebey and Gold 1988) because the latter affects the rate of drug delivery into the brain (Perez-Reyes et al 1982). For example, whereas a 50% DAT blockade is associated with a “high” when MP is injected IV and leads to peak concentrations in brain 8–15 min after its administration, this same level of blockade induced by oral MP, which leads to peak concentrations in brain 60–120 min after its administration, does not produce a “high” (Volkow et al 1998). Though much less investigated, the rate of clearance of the drug from brain also affects its reinforcing effects and abuse potential because it is likely to limit the frequency at which the drugs can be self-administered. In this respect, DAT-blocking drugs with very fast rates of clearance, such as cocaine, are much more likely to promote frequent self-administration, because the reinforcing effects are associated with the rapid blockade of more than 50% of the DAT (Volkow et al 1995a). If DAT drugs are blocking more than 50% of DAT with a single administration, then slow pharmacokinetics will lead to DAT saturation with repeated frequent administration and will eventually interfere with further administration (Volkow et al 1995a). Thus, even if radafaxine were to be injected IV at doses large enough to induce more than 50% DAT blockade, its long DAT blockade is likely to interfere with its repeated administration. We also postulate that fluctuating versus steady-state drug concentrations in brain will affect the drug’s reinforcing effects. We postulate this because animal studies have shown that the rate at which animals self-administer DAT-blocking or DA- releasing drugs, such as cocaine and , is predicted by the downward slope of DA that follows the drug-induced increases in nucleus accumbens (Ranaldi et al 1999; Wise et al 1995). Although such studies have not been done in humans, studies in human cocaine abusers have shown that craving increases as the “high” induced by cocaine decreases (Breiter et al 1997). Because we have shown that the “high” in humans from IV MP is associated with DA increases (Volkow et al 1999c), it is also possible that in humans the craving is triggered by the decreases in DA that follow the peak after drug administration. Also, because the reinforcing effects of DAT blockers are asso- ciated with the fast change in DA (⌬ in DA concentration) Figure 7. Cardiovascular measures before and at different times after ad- (Volkow et al 1999c), we postulate that drugs that have long- ministration of radafaxine. Values correspond to means and SDs. None of the effects were significant. lasting steady-state concentrations are less likely to induce drug craving than those that do not. The temporal course of DAT blockade corresponded well slow pharmacokinetics, and its steady-state DAT blockade sug- with the pharmacokinetics of the drug in plasma, which peaked gest that radafaxine is likely to have minimal or no reinforcing at approximately 4–6 hours after drug intake and had a clearance potential and is therefore unlikely to be abused. Specifically, we rate at 24 hours of 13%. Moreover, the concentration of radafax- have shown that for DAT blockers to be perceived as reinforcing ine in plasma predicted the peak levels of DAT blockade. At the they need to block more than 50% of DAT (Volkow et al 1997, estimated peak radafaxine in plasma, concentrations of 89.7 1999a, 1999b). We have also shown that in addition to achieving ng/mL that also occurred at approximately 4–6 hours after drug a Ͼ50% DAT blockade, the rate of DAT blockade has to be fast intake, the induced DAT blockade corresponds to 19%. The for the reinforcing effects to occur (Volkow et al 1995a, 1998, significant correlation between plasma and DAT blockade sug- 2000). Thus, when given orally, 50% DAT blockade is not gests that plasma concentration of radafaxine is a good surrogate perceived as reinforcing, but IV administration is. It therefore marker for monitoring radafaxine delivery to the brain. seems that it is the rate of DAT blockade (percent blockade per In this study, a 40-mg dose of oral radafaxine was devoid of unit of time) that is relevant for the reinforcing effects to occur any observable behavioral or cardiovascular effects. The lack of (Volkow and Swanson 2003). these adverse effects with 20% DAT blockade is not surprising It is known that the rapidity and abruptness of drug delivery because we had previously shown that drug doses that induced to the brain affect the reinforcing effects of drugs; the shorter the DAT blockade of less than 40% had behavioral and cardiovas- interval between intake and perceived effects of the drug, the cular effects that did not differ from those of placebo (Volkow et greater the reinforcing effects of the drug (Balster and Schuster al 1997). We postulate that this is due to an excess of DAT, such 1973; Oldendorf 1992; Samaha et al 2004). This in turn provides that under baseline conditions (without stimulation) the excess www.elsevier.com/locate/biopsych N.D. Volkow et al BIOL PSYCHIATRY 2005;57:640–646 645

DAT can compensate for the low level of blockade. In fact, a relevant variable for drug reinforcement is not just DA increases but simulation model to predict the relationship between DAT “fast DA increases” (Volkow and Swanson 2003), even if radafaxine blockade and changes in DA predicted that with stable DA were to be injected (at large enough doses), and assuming that it release, 40% DAT blockade would not change extracellular DA would have reinforcing effects, its very low clearance from brain (Gatley et al 1997). However, low/lower DAT blockade might be would result in saturation of DAT, which would interfere with its relevant during phasic DA cell firing, which might amplify the frequent repeated administration (DAT would be occupied, so magnitude and duration of extracellular DA. Failure of radafaxine subsequent doses would not be able to further change DA). to induce a “high,” change “mood,” or induce desire for more In this study, we used subjective self-reports to assess the drug is concordant with its lack of reinforcing effects. Because reinforcing effects of radafaxine in healthy subjects. Although we descriptors of drug effects, such as the “high,” are the best did not use other tests for reinforcing effects, such as choice predictor of drug self-administration in humans (Fischman and procedures, self-reports of drug effects have been shown to be Foltin 1991), this also predicts that the abuse and diversion of reliable and to predict drug self-administration in humans (Fis- radafaxine would be unlikely. chman and Foltin 1991). In addition, this study was done in In interpreting the findings from this study, it is important to non–drug abusers, so we cannot rule out the possibility that drug comment on [11C]cocaine as a tracer for the DAT, as well as on abusers might have perceived radafaxine as reinforcing. This is the model used in this study to assess DAT occupancy (Bmax/Kd unlikely, however, because comparison of the rewarding effects ϭ Ϫ DVSTR/DVCB 1). Although cocaine binds to the DAT, the of MP is not greater in addicted than in nonaddicted subjects norepinephrine transporter, and the (Ritz et (Volkow et al 1997). Moreover, when cocaine is given to cocaine al 1987), the binding of [11C]cocaine in the striatum as measured abusers at doses that do not induce more than 50% DAT by PET has been shown to represent only the DAT (Volkow et al blockade, they do not recognize the drug as different from 1995b). More specifically, studies in humans have shown that placebo (Volkow et al 1997). Thus, it is unlikely that a dose that striatal [11C]cocaine binding is insensitive to pretreatment with control subjects do not differentiate from placebo would be (Fowler et al 1989), demonstrating that binding to recognized as reinforcing by cocaine abusers. the norepinephrine transporter is negligible. Similarly, studies in A final limitation concerns the statistical analysis used to the baboon have shown that [11C]cocaine binding in the striatum assess the correlation analysis, owing to the relatively small is not affected by drugs that bind to the serotonin transporter sample size (n ϭ 8) and the fact that measures were obtained for ( and ) (Volkow et al 1995b). Dopamine the same subjects at different time points, so that these observa- transporter occupancy is assessed by graphic analysis of [11C]co- tions are not independent from one another, which violates caine time–activity data, which allows the measurement of the statistical assumptions in modeling. Nonetheless, the results distribution volume (DV) (Logan et al 1990, 1996). The DV is an predicted from the modeling of the data are in agreement with “equilibrium” measure representing the ratio of tissue radioactiv- the results obtained with the PET measures and thus provide ity (metabolite corrected) to plasma radioactivity. Because it is an preliminary evidence that plasma measures might serve as an equilibrium property, it does not depend on blood flow. The DV estimate for brain DAT occupancy. calculated for [11C]cocaine with the graphic method has been The findings from this study can only “predict” but do not shown to be the same as that obtained from a compartmental “indicate” that the radafaxine will not be reinforcing. Indirect model approach and to be a stable parameter that does not evidence that this prediction is correct is given by the infre- change with time (Logan et al 1990, 1996). Furthermore, esti- quency of abuse of (Margolin et al 1995; Miller and mates of DAT occupancy have been shown to be reproducible Griffith 1983; Peck et al 1979; Rush et al 1998), which has (Volkow et al 1997) and to lead to similar measures whether one pharmacologic properties similar to those of radafaxine. In fact uses [11C]cocaine or [11C]d-threo methylphenidate as the DAT there is no evidence of abuse or diversion of bupropion when radioligand (Fowler et al 1998). given to cocaine abusers with attention-deficit/hyperactivity dis- Limitations of this study include the fact that the studies were order (Levin et al 2002; Montoya et al 2002). Nonetheless, future done with acute and not chronic administration. The long-lasting epidemiologic data will allow us to evaluate whether this pre- DAT blockade of DAT with radafaxine predicts that chronic diction is correct. administration should result in higher levels of DAT occupancy. In summary, we have shown significant though relatively low Even if greater DAT blockade is achieved with chronic adminis- levels of DAT blockade in the human brain after oral adminis- tration, however, this does not affect the interpretation that its tration of radafaxine. Its relatively low potency in blocking DAT, low potency predicts poor reinforcing effects because the in- its slow rate of DAT blockade, and its long-lasting blockade creases would occur over a long period. Also, because the route predict a very poor reinforcing profile for this drug. This study is of administration affects the reinforcing effects of drugs (Verebey the first in which PET technology was used to investigate the and Gold 1988), we cannot rule out that if radafaxine were to be potency and pharmacokinetics of a new at its injected IV it would not be perceived as reinforcing. Indeed, PET molecular target in humans, to predict its potential reinforcing studies of the relationship between the DA increases induced by effects and hence abuse liability. MP and its reinforcing effects showed that for equivalent DA This research was carried out at Brookhaven National Labo- increases, oral MP was not reinforcing, whereas IV MP was ratory with the support of GlaxoSmithKline. (Volkow et al 1999c, 2001). In extrapolating data from MP to We thank Noel Netusil and Victor Garza for their contribu- radafaxine, however, one has to consider that MP is significantly tions to this work. more potent at blocking DAT than radafaxine (ED50 or dose required for MP to block 50% of DAT is .23 mg/kg) (Volkow et al 1998). Because preclinical studies have shown that DAT Balster RL, Schuster CR (1973): Fixed-interval schedule of cocaine reinforce- ment: Effect of dose and infusion duration. J Exp Anal Behav 20:119–129. inhibitors with low DAT occupancies do not maintain drug Blazer DG, Kessler RC, McGonagle KA, Swartz MS (1994): The prevalence and self-administration, it is not evident that even if injected radafaxine distribution of major depression in a national community sample: The would be self-administered (Lindsey et al 2004). Also, because the National Comorbidity Survey. Am J Psychiatry 151:979–986.

www.elsevier.com/locate/biopsych 646 BIOL PSYCHIATRY 2005;57:640–646 N.D. Volkow et al

Breiter HC, Gollub RL, Weisskoff RM, Kennedy DN, Makris N, Berke JD, et al Perez-Reyes M, Di Guiseppi S, Ondrusek G, Jeffcoat AR, Cook CE (1982): (1997): Acute effects of cocaine on human brain activity and emotion. Free-base cocaine smoking. Clin Pharmacol Ther 32:459–465. Neuron 19:591–611. Ranaldi R, Pocock D, Zereik R, Wise RA (1999): Dopamine fluctuations in the Damaj MI, Carroll FI, Eaton JB, Navarro HA, Blough BE, Mirza S, et al (2004): nucleus accumbens during maintenance, extinction, and reinstatement Enantioselective effects of hydroxy metabolites of bupropion on behav- of intravenous D-amphetamine self-administration. J Neurosci 19:4102– ior and on function of monoamine transporters and nicotinic receptors. 4109. Mol Pharmacol 66:675–682. Ritz MC, Lamb RJ, Goldberg SR, Kuhar MJ (1987): Cocaine receptors on Fischman MW, Foltin RW (1991): Utility of subjective-effects measurements dopamine transporters are related to self-administration of cocaine. in assessing abuse liability of drugs in humans. Br J Addict 86:1563–1570. Science 237:1219–1223. Fowler JS, Volkow ND, Logan J, Gatley SJ, Pappas N, King P, et al (1998): Rush CR, Kollins SH, Pazzaglia PJ (1998): Discriminative-stimulus and partic- Measuring dopamine transporter occupancy by cocaine in vivo: Radio- ipant-rated effects of methylphenidate, bupropion, and triazolam in tracer considerations. Synapse 28:111–116. d-amphetamine-trained humans. Exp Clin Psychopharmacol 6:32–44. Fowler JS, Volkow ND, Wolf AP, Dewey SL, Schlyer DJ, MacGregor RR, et al Samaha AN, Mallet N, Ferguson SM, Gonon F, Robinson TE (2004): The rate of (1989): Mapping cocaine binding sites in human and baboon brain in cocaine administration alters gene regulation and behavioral plasticity: vivo. Synapse 4:371–377. Implications for addiction. J Neurosci 24:6362–6370. Gatley SJ, Volkow ND, Gifford AN, Ding YS, Logan J, Wang GJ (1997): Model Sanchez C, Hyttel J (1999): Comparison of the effects of antidepressants and for estimating dopamine transporter occupancy and subsequent in- their metabolites on of biogenic amines and on receptor bind- creases in synaptic dopamine using positron emission tomography and ing. Cell Mol Neurobiol 19:467–489. carbon-11-labeled cocaine. Biochem Pharmacol 53:43–52. Verebey K, Gold MS (1988): From leaves to crack: The effects of dose and Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M, Eshleman S, et al routes of administration in abuse liability. Psychiatry Ann 18:513–520. (1994): Lifetime and 12-month prevalence of DSM-III-R psychiatric disor- Volkow ND, Ding YS, Fowler JS, Wang GJ, Logan J, Gatley JS, et al (1995a): Is ders in the United States. Results from the National Comorbidity Survey. methylphenidate like cocaine? Studies on their pharmacokinetics and Arch Gen Psychiatry 51:8–19. distribution in the human brain. Arch Gen Psychiatry 52:456–463. Lam RW, Wan DD, Cohen NL, Kennedy SH (2002): Combining antidepres- Volkow ND, Fowler JS, Logan J, Gatley SJ, Dewey SL, MacGregor RR, et al sants for treatment-resistant depression: A review. J Clin Psychiatry 63: (1995b): Carbon-11-cocaine binding compared at subpharmacological 685–693. and pharmacological doses: A PET study. J Nucl Med 36:1289–1297. Volkow ND, Swanson JM (2003): Variables that affect the clinical use and Levin FR, Evans SM, McDowell DM, Brooks DJ, Nunes E (2002): Bupropion abuse of methylphenidate in the treatment of ADHD. AM J Psychiatry treatment for cocaine abuse and adult attention-deficit/hyperactivity 160:1909–1918. disorder. J Addict Dis 21:1–16. Volkow ND, Wang GJ, Fischman MW, Foltin R, Fowler JS, Franceschi D, et al Lindsey KP, Wilcox KM, Votaw JR, Goodman MM, Plisson C, Carroll FI, et al (2000): Effects of route of administration on cocaine induced dopamine (2004): Effects of dopamine transporter inhibitors on cocaine self-ad- transporter blockade in the human brain. Life Sci 67:1507–1515. ministration in emission tomography neuroimaging. J Pharmacol Exp Volkow ND, Wang GJ, Fischman MW, Foltin RW, Fowler JS, Abumrad NN, et al Ther 309:959–969. (1997): Relationship between subjective effects of cocaine and dopa- Logan J, Fowler JS, Volkow ND, Wang GJ, Ding YS, Alexoff DL (1996): Distri- mine transporter occupancy. Nature 386: 827–830. bution volume ratios without blood sampling from graphical analysis of Volkow ND, Wang GJ, Fowler JS, Fischman M, Foltin R, Abumrad NN, et al PET data. J Cereb Blood Flow Metab 16:834–840. (1999a): Methylphenidate and cocaine have a similar in vivo potency to Logan J, Fowler JS, Volkow ND, Wolf AP, Dewey SL, Schlyer DJ, et al (1990): block dopamine transporters in the human brain. Life Sci 65: L7–L12. Graphical analysis of reversible radioligand binding from time-activity Volkow ND, Wang GJ, Fowler JS, Gatley SJ, Logan J, Ding YS, et al (1999b): measurements applied to [N-11C-methyl]-(-)-cocaine PET studies in hu- Blockade of striatal dopamine transporters by intravenous methylpheni- man subjects. J Cereb Blood Flow Metab 10:740–747. date is not sufficient to induce self-reports of “high”. J Pharmacol Exp Ther Logan J, Volkow ND, Fowler JS, Wang GJ, Dewey SL, MacGregor R, et al 288:14–20. (1994): Effects of blood flow on [11C]raclopride binding in the brain: Volkow ND, Wang GJ, Fowler JS, Gatley SJ, Logan J, Ding YS, et al (1998): model simulations and kinetic analysis of PET data. J Cereb Blood Flow Dopamine transporter occupancies in the human brain induced by ther- Metab 14:995–1010. apeutic doses of oral methylphenidate. Am J Psychiatry 155:1325–1331. Margolin A, Kosten TR, Avants SK, Wilkins J, Ling W, Beckson M, et al (1995): Volkow ND, Wang GJ, Fowler JS, Logan J, Gatley SJ, Wong C, et al (1999c): A multicenter trial of bupropion for cocaine dependence in - Reinforcing effects of psychostimulants in humans are associated with maintained patients. Drug Depend 40:125–131. increases in brain dopamine and occupancy of D(2) receptors. J Pharma- Miller L, Griffith J (1983): A comparison of bupropion, , col Exp Ther 291:409–415. and placebo in mixed-substance abusers. Psychopharmacology (Berl) Volkow ND, Wang G, Fowler JS, Logan J, Gerasimov M, Maynard L, et al 80:199–205. (2001): Therapeutic doses of oral methylphenidate significantly increase Montoya ID, Preston KL, Rothman R, Gorelick DA (2002): Open-label pilot extracellular dopamine in the human brain. J Neurosci 21:RC121. study of bupropion plus bromocriptine for treatment of cocaine depen- Wang GJ, Volkow ND, Hitzemann RJ, Wong C, Angrist B, Burr G, et al (1997): dence. AM J Drug and Alcohol Abuse 28:189–96. Behavioral and cardiovascular effects of intravenous methylphenidate Oldendorf WH (1992): Some relationships between addiction and drug de- in normal subjects and cocaine abusers. Eur Addict Res 3:49–54. livery to the brain. NIDA Res Monogr 120:13–25. Wise RA, Newton P, Leeb K, Burnette B, Pocock D, Justice JB Jr (1995): Peck AW, Bye CE, Clubley M, Henson T, Riddington C (1979): A comparison of Fluctuations in nucleus accumbens dopamine concentration during in- bupropion hydrochloride with dexamphetamine and in travenous cocaine self-administration in rats. Psychopharmacology (Berl) healthy subjects. Br J Clin Pharmacol 7:469–478. 120:10–20.

www.elsevier.com/locate/biopsych